Efficient Multi-Objective Optimization of an Automotive Crush Rail Under Axial and Oblique Impact

Abstract

Oblique impact conditions are of increasing research importance, and this kind of study is becoming more common in the crashworthiness literature. Optimization strategies, such as the Response Surface Methodology (RSM), are commonly used in a virtual design environment to investigate various loading conditions and configurations. However, the typical usage of the RSM in the crashworthiness literature does not take advantage of many recent enhancements to the RSM. This thesis presents a multi-objective shape optimization study of a multi-cellular aluminum extrusion subjected to dynamic axial and oblique (20° angled) impact loading conditions. Finite element models are developed, validated with experimental data, parameterized, and used to exhaust the design space using a complete factorial design of experiments. An analysis of different energy absorption characteristics, such as axial and oblique mean crush force, crush efficiency, and oblique impact coefficient, are presented. A multi-objective optimization problem is defined to generate new solutions that dominate a baseline profile in all aspects of performance, which produced a geometry with at least 6-17% improvements in energy absorption. The parametric study is used to define the “true” Pareto front, which is the set of non-dominated designs and the solution to the optimization problem. The RSM is used to perform the multi-objective optimization analysis. Feedforward neural networks and non-dominated sorting genetic algorithms (NSGA-II) are used as the metamodel and optimizer, respectively. This thesis research utilizes the Adaptive Surrogate-Assisted (ASA) - RSM, which is an advanced RSM approach, to perform the optimization analysis and demonstrate performance enhancements. The traditional RSM approach is compared to ASA-RSM in their speed and ability to correctly identify the actual Pareto front. The traditional RSM approach required approximately 63% of the entire domain to accurately identify the 94% of true Pareto front. The ASA-RSM was able to identify 87% of the true Pareto front using 25% of the entire domain. In some instances, the ASA-RSM was able to correctly identify the true Pareto front by using 33% of the domain. The traditional RSM approach was only able to correctly identify 10% of the true Pareto front when sampling less than 15% of the domain or using traditional stopping criteria. Depending on the configuration, the ASA-RSM achieved 40-80% accuracy in predicting the Pareto front when sampling up to 15% of the domain. This study provides a comprehensive understanding of the performance gains of multi-objective optimization in the application of structural crashworthiness considering various loading conditions.